Professional Documents
Culture Documents
Learning Outcomes
This chapter builds upon and extends the basic principles of control systems that are dealt with
in Control Principles 1. On completion you should be able to:
1 Appreciate the different forms of system element, and analogous elements.
2 Identify a range of first-order systems, and describe their response to step disturbance inputs.
3 Distinguish between first-order and second-order systems.
4 Describe the response of second-order systems to both step inputs and forcing functions,
particularly with reference to system damping.
5 Explain the effect of mechanical resonance, and methods of reducing its effects.
6 Understand the concepts of proportional, differential and integral control strategies.
⎛1 ⎞ ⎛1 ⎞ ⎛ d q ⎞⎟
Electrical inductor ⎜⎜ LI 2 ⎟⎟⎟ capacitor ⎜⎜ CV 2 ⎟⎟⎟ resistor ⎜⎜ R ⎟
⎜⎝ 2 ⎠ ⎜⎝ 2 ⎠ ⎜⎝ d t ⎟⎠
122
Control Principles 2 123
Both the mass and the inductor store energy in a form associated with
movement. In the case of the mass, this will be kinetic energy by virtue
of its velocity. For the inductor, the energy is stored in the magnetic
field produced by the current flowing. Remember, v ds/dt metre per
second; and I dq/dt coulomb/second (amp), i.e. the current is the rate
at which charge is displaced.
Both the spring and the capacitor store energy by virtue of position
or potential—that is, potential energy. For example, the further the
spring is displaced the more energy it will store. Similarly, the more
charge placed on the plates of a capacitor the larger its p.d., hence more
energy stored.
Both the damper and the resistor have a ‘slowing down’ effect in their
respective systems. As a result they cause heat energy to be dissipated.
voltage, V force, F
charge, Q displacement, s
current, I vel o city, v
resistance, R damper coefficient, R
inductance, L m ass, m
capacitance, C reciprocal of spring stiffness, 1/ k
2 First-order Systems
A first-order system is one which contains only one of the two possible
types of energy storage element, in combination with the associated
energy dissipative element. For the elements so far considered, the
following combinations form first-order systems:
Mass-damper; spring-damper; inductor-resistor; and capacitor-
resistor. The last two combinations we have already studied in detail
when considering d.c. transients in the book, Further Electrical and
Electronic Principles, ISBN 9780750687478 (Chapter 8). In this
situation the systems (circuits) were subjected to a step input, by
connecting to or disconnecting from a d.c. supply. In each case it was
found that the system response followed an exponential law. Let us
review the salient points regarding the C-R and L-R series circuits,
and see how these results can be translated in terms of the response of
analogous mechanical systems.
124 Control Principles 2
R C q
i
VR VC
Fig. 1
E vR vC volt
Rdq q
since vR Ri ; and vC
dt C
Rdq q
then, E voolt.................[1]
dt C
S
k
F2
R
F
F1
Fig. 2
If the switch in the C-R circuit is now returned to its original position
then the capacitor will discharge exponentially. Similarly, if the applied
force is now removed from the spring, it will return to its normal
length. This displacement will also be exponential in form.
m
F1
F2
R
Fig. 3
126 Control Principles 2
F F1 F2 newton
where, F1 mass acceleration
dv
ma newton m neewton
dt
and F2 Rv newton
dv
therefore, F m Rv newton
dt
F
Hence, v (1 et/( ) ) metre/second;
R
m
and seconds
R
di
VL Ri volt
dt
dx
YA Bx
dt
Y
x (1 et/ ) for exponential growth
B
Y
x et/ for exponent i al decay;
B
A
and seconds
B
Control Principles 2 127
Worked Example 1
Q A spring-damper system as shown in Fig. 4 is subjected to a force of 25 N. The spring stiffness is 200 N/
m and the damper coefficient is 600 Ns/m. If this force is suddenly removed, determine (a) the system
time constant, (b) the time taken for the system to move 50 mm from its initial starting point, and (c)
the time taken for the spring to return to its ‘normal’ uncompressed length.
A
F 25 N; k 200 N/m; R 600 Ns/m
F 25 N
Fig. 4
R 600
(a) seconds
k 200
therefore 3 s Ans
(b) In order to determine details of how the spring will return to its unstressed
length, we firstly have to determine the distance the applied force will
initially displace it.
Initial displacement,
F 25
si metre 125 mm
k 200
Comparing this system with its electrical analogue, we can say that this
displacement corresponds to the initial charge placed on a capacitor. When
the force is released it will be equivalent to starting the discharge process
for the capacitor. Thus the displacement for which the time is required to be
calculated involves a decaying exponential function.
s si et/ metre
s
et/
si
t s
ᐉn
si
t 50
ᐉn 0.9163
3 125
hence t 2.75 s Ans
(c) The system will take approximately five time constants to reach its new
steady state, so time taken for spring to unstretch 15 s Ans
128 Control Principles 2
s
k
m PEmax
(a)
KEzero
k
(b) m PEzero
KEmax
s
k
(c) m PEmax
KEzero
datum
Fig. 5
will begin to stretch. Thus the mass will now be transferring energy
back to the spring. Figure 5(c) shows the instant that the mass comes
to rest, at the system maximum displacement on the other side of the
datum. At this point all of the energy will once more be stored in the
spring. The whole sequence will then be repeated, and the system will
oscillate back and forth with a constant amplitude of s metre. Since no
energy is lost from the system, then these oscillations would continue
indefinitely—perpetual motion! In exactly the same way as was used in
the first-order systems, we can deduce the system differential equation
for this system by equating the forces involved, as follows.
s(m)
S
0
t(s)
S
Fig. 6
130 Control Principles 2
k
n rad/s
m (1)
1 k
so, fn hertz
2 m (2)
‘1’ ‘2’
E L
C
Fig. 7
It may occur to you that these equations seem vaguely familiar. That
is as it should be, since the equivalent electrical system is the L-C
circuit shown in Fig. 7. In this circuit, the capacitor would initially be
charged with the switch in position ‘1’. When the switch is moved to
position ‘2’ the energy stored in C will drive a current through L, thus
transferring energy to the magnetic field of the inductor. The result will
be a continuous oscillation of current and energy between the capacitor
and inductor (assuming zero resistance). The frequency of oscillation
would be:
1 1
n rad/s; or fn hertz
LC 2 LC
You should recognise the above equations as those for the resonant
frequency for a series L-C circuit, or, for a parallel resonant circuit
where the resistance is zero. Bearing in mind that according to our
analogies C 1/k, then it may be seen that the resonant or natural
frequency equation for the electrical circuit corresponds directly with
that for the mechanical system shown.
the three elements are shown as F1 to F3, and the system differential
equation will be:
F F1 F2 F3 newton
m d s2 Rds
2
k
F3
F1
m F
F2
R Si
Fig. 8
S(m)
Si
exponential
‘envelope’
0
t(s)
Fig. 9
more rapidly will the oscillations die away. Indeed, if the damping is
increased sufficiently, a point will be reached where the oscillations
are just prevented from occurring. This degree of damping is known as
critical damping Rcrit. Under this condition the response of the system
will be approximately exponential.
Although the system will have a specific value of damping coefficient
R, the overall damping is normally referred to in terms of the system
damping ratio (Greek letter zeta). Damping ratio is defined as the
ratio of the damping present to that required to just prevent oscillations.
In the form of an equation this is:
R
(3)
Rcrit
fd fn 1 2 hertz (4)
The equivalent electrical circuit is shown in Fig. 10, and the relevant
system equations are listed below.
E vL vR vC volt
d2 q dq q
so, E L R volt
dt 2 dt C
R L C
i VR VL VC
‘2’ ‘1’
E
Fig. 10
1 1
n rad/s; fn hertz
LC 2 LC
d n 1 2 rad/s;
fd fn 1 2 hertz
Control Principles 2 133
In this circuit, when the switch is moved from position ‘1’ to ‘2’ the
charge in the system will increase from zero to its final steady value,
whilst the circuit current will decrease from its initial value, to zero.
The graphs of these variations are illustrated in Fig. 11. Returning the
switch to position ‘1’ would be equivalent to releasing the mass-spring-
damper as previously described.
q (C)
0 t (s)
i (A)
o
0
t (s)
Fig. 11
E Lq Rq q/ C
where q d 2 q/dt 2 and q dq /dt
E Rq q
q
L L LC
The coefficient of the last term is 1/LC n2. It is found that the
coefficient of the second term, R/L 2n. Thus, if any second-order
system has an equation of the form
Y ax bx cx
Y bx cx
then
x
a a a
Y
and
x 2 n x 2n x
a (5)
134 Control Principles 2
1.6
ζ 0.3
1.4
ζ 0.5
1.2
ζ 0.7
1.0
θo /θi
0.8
ζ 1.3
ζ1
0.6
0.4
0.2
0 1 2 3 4 5 6
ζωnt
Fig. 12
Worked Example 2
Q A linear mechanical system has a total mass of 15 kg, an effective spring stiffness of 10 N/m, and a
damping coefficient of 20 Ns/m. If this system is subjected to a sudden applied force of 25 N, write
down the system differential equation and hence determine (a) the natural frequency, and (b) the
system damping ratio.
A
m 15 kg; k 10 N/m; R 20 Ns/m; F 25 N
(a) hence, 25 s 20 s 10 s
15 15 15
and comparing coefficients with the general equation (5)
10
n 0.816 rad/s
15
0.816
so fn 0.13 Hz Ans
2
Control Principles 2 135
20
(b) 2n 1.3333
15
1.333
therefore n 0.817 Ans
2 0.816
s.h.m
m
Fig. 13
5.0
4.5
4.0
3.5
3.0 ζ 0.1
So /Si
2.5 ζ 0.2
2.0
ζ 0.3
1.5
ζ 0.4
1.0
ζ 0.5
ζ 0.7
0.5 ζ1
ζ2
Fig. 14
f/fn
0 0.5 1.0 1.5 2.0 2.5
ξ 0.1
ξ 0.2
45 ξ 0.5
ξ1
ξ2
(degrees)
angle φ
90
135
180
Fig. 15
then the system damping must be increased. This will have the effect of
reducing the amplitude of oscillations. From the amplitude/frequency
response curve it may be seen that if 0.7, then the displacement
of the mass will not exceed that of the input. Also, at frequencies in
excess of 1.5fo the displacement of the mass becomes progressively
less and less. A classic example of these effects may be experienced
when driving a car in which the roadwheels are out of balance. In this
case it will be found that at one particular speed the vibrations due to
the imbalance become very noticeable. If the speed is either increased
or decreased from this value it will be found that the vibrations become
much weaker, or even disappear altogether.
Proportional control
This control strategy is called proportional control because the
feedback signal is directly proportional to the angular position of the
output.
138 Control Principles 2
Vi error
θi signal Reference
A1 A2 supply
Vo P2
P1
Reference Constant
n
supply field M LOAD
supply 1
θo
Vo
A3
Positional (proportional) feedback
Fig. 16
P1 and P2
These are the input and output potentiometers (transducers) which
provide the electrical signals (Vi and Vo) that are proportional to the
input or demanded displacement and the actual output displacement, i
and θo respectively.
A1
This is the error detector, in the form of a summing operational amplifier.
The output of this device will be the error signal, equal to ViVo.
A2
This is a power or servo amplifier which provides a much amplified
error signal to the drive motor.
M
This is a separately excited d.c. driving motor.
n
This is a step-down gearbox to provide the large torque required by
the load.
A3
This is an inverting operational amplifier to ensure that the feedback
signal Vo is in opposition to the input Vi. This ensures that negative
feedback is applied.
In addition to the mass of the load are the masses of motor armature,
drive shafts, gearbox etc. Therefore the total mass, and hence inertia,
of the system will be large. The ‘stiffness’ of the system is contributed
Control Principles 2 139
by the drive shafts and gearbox, and the amplification in the system
will have a similar effect to mechanical stiffness. In order to minimise
any steady state error the system gain (amplification) needs to be high.
On the other hand, the friction in the system will be minimal due to
the use of high quality bearings, lubrication etc. The whole system will
therefore be a severely underdamped second-order system which, due
to the high gain, will be very responsive to demanded changes. Thus,
when a step demand input is applied, the system will respond rapidly
with consequent violent oscillations of the antenna about its new
position before finally settling. This oscillatory response is obviously
undesirable, especially since large masses are involved. The solution is
to slow down the system by increasing the damping ratio.
This effect could be achieved by increasing the friction in the system,
perhaps by the use of some form of braking. However, increasing
viscous friction is to be avoided because it wastes energy, it increases
steady state error, and it was initially designed to be minimal anyway!
Thus the slowing down effect of viscous friction needs to be simulated,
without the other adverse effects. This may be achieved by means of
derivative feedback as follows.
Derivative control
The original positional (proportional) feedback must be retained since
this forms the basic closed-loop control which ensures that the system
is error-actuated. The derivative feedback may be achieved by several
means, one of which is by the use of a tachogenerator mounted on the
motor driveshaft as shown in Fig. 17.
The output of the tachogenerator is directly proportional to the speed
at which it rotates, so it is proportional to the rate of change of output
position, i.e. Vd dθo /dt, so the tacho voltage is the derivative of the
output position. This feedback signal is also of the opposite polarity
Vi Reference
θi A2 supply
Vo A1
P2
Vd
P1
Reference Constant n
supply field M T.G. LOAD
1 θo
supply
P3
Derivative (velocity) feedback
Vo
A3
Proportional (position) feedback
Fig. 17
140 Control Principles 2
to the demand signal, so the total error signal will now be given by
Vi(VoVd) volt. The major advantage of tachogenerator feedback
is that it will be at its greatest when the output shaft is travelling at
its fastest (through the oscillatory ‘cross-over’ points on the response
graph), and will be zero when the shaft is stationary. This means that
the damping effect of this feedback is present only when it is required
and will, unlike friction, not affect the steady state error in the shaft
position. In addition, the degree of damping produced can very simply
be adjusted by means of the associated potentiometer, P3.
Although the use of a tachogenerator would seem to be the ideal
solution to the problem of oscillatory instability in a positional control
system, it can have an adverse effect in a velocity control system, or a
positional control system subjected to a ramp input demand.
Velocity lag
A ramp input is simply a step velocity input, as illustrated in Figs. 18(a)
and (b). This type of input would apply to a positional control system
when it is required to rotate at some constant velocity. For example,
this would apply to a radar antenna when it is required for the antenna
to lock on to and follow a ‘target’. Since the system has to be error-
actuated, then to keep the motor driving there has to be some difference
between o and i, to provide the necessary error signal. Consider the
antenna rotating at the appropriate speed to match that of the ‘target’
(o i). In order to provide the error signal to maintain this velocity
there will be an angular misalignment between the actual instantaneous
position of the antenna and the desired position. This means that the
antenna angular position will be ‘lagging behind’ the target. This
angular difference is called the velocity lag, and from Figs. 19(a) and
(b) it may be seen that velocity lag (steady state error) increases with
damping. Hence if a tachogenerator is used, the damping signal that it
produces whilst rotating will increase the velocity lag. This problem
may be avoided if the velocity feedback is produced by passing the
error signal through either a passive C-R differentiating network, or a
differentiating operational amplifier. The differentiator and the error
detector usually form parts of an element called a controller. Thus a
demand demand
θi dθi
ωi
dt
t (s) t (s)
ramp displacement step velocity
(a) (b)
Fig. 18
Control Principles 2 141
θ θ
θi θi
θo
θo θss
θss
t (s) t (s)
(a) overdamped (b) underdamped
Fig. 19
Integral control
The differentiating network mentioned above is satisfactory for low-
power systems in which the load on the motor is relatively small
and constant. However, where there is significant ‘stiction’ (static
or coulomb friction, independent of velocity) in the system, then
steady state errors can still be significant, since stiction increases the
dead-band. As the load approaches the desired alignment position
(o i), the error voltage becomes progressively smaller. Figure
20 illustrates this point and shows the dead-band. In this context, the
dotted lines represent the minimum error voltage required to overcome
the stiction. Thus, due to its inertia, the load could come to rest at
any position between points (a) and (b). The system dead-band could
be reduced by increasing the system gain, but too much gain tends
to produce instability, in the form of oscillations. The solution to the
problem is the use of integral feedback, either in the form of a passive
C-R integrating network or an integrating operational amplifier. In
either case, a p.d. is built up across the capacitor whilst the load is
approaching alignment. This p.d. is sufficient to supplement the error
signal to the extent that sufficient error voltage will now be present
to overcome the dead-band. With careful adjustment it is possible to
eliminate the steady state error. The arrangement for a three-term or
PID controller is shown in Fig. 21.
error
0 dead-band
(a) (b)
Fig. 20
142 Control Principles 2
Vi
error
Vi Vo
P
to servo
Vo amplifier
Fig. 21
digital analogue
Computer/ error error
DAC Controller System
µprocesser
output
θo
analogue
digital feedback
feedback Vo
ADC Transducer
Fig. 22
Control Principles 2 143
Summary of Equations
First-order systems:
dx
General system equation, Y A Bx
dt
A
Time constant, second
B
Y
Instantaneous value for growth, x (1 et/ )
B
Y t/
Instantaneous value for decay, x e
B
Ideal second-order:
General equation (SHM), Y ax bx
1 b
Natural frequency, fn hertz
2 a
Practical second-order:
Y bx cx
General equation,
x x 2ζvn x vn2 x
a a a
1 c
Natural frequency, vn rad/second
2π a
R
Damping ratio, ζ
Rcrit
Assignment Questions
1 Deduce the electrical analogues for (b) the time constant, and (c) the final
torque, inertia, angular velocity, angular velocity.
displacement, and a torsion bar.
7 A system comprises a 2.4 kg mass, a spring
2 A spring of stiffness 1.5 kN/m is attached of stiffness 9.6 N/m, a damper of coefficient
to a mass of 20 kg. What will be the natural 3.2 Ns/m, and is subjected to a step input force
frequency of oscillation of this system when of 0.5 N. Determine the system differential
disturbed? equation and hence, or otherwise, calculate
3 A mass-spring system has a natural frequency (a) the natural frequency, (b) the damping
of 1.5 Hz. If the mass is 4 kg determine the ratio, (c) the damped frequency, (d) the final
value of spring stiffness. displacement, and (e) the value of damper
coefficient that will just prevent oscillations.
4 A mass-spring system has a natural frequency
of 0.5 Hz. If it is then damped by a device 8 Explain why resonance in mechanical
having a damping ratio of 0.65, determine the systems is normally avoided, and if it cannot
damped frequency of oscillation. be avoided, how the adverse effects are
minimised.
5 Explain the difference between first-order and
second-order systems, quoting examples of 9 Explain what a tachogenerator is and why
each. it may be employed in a positional control
system.
6 A mass of 63 kg is attached to a damper of
coefficient 30 Ns/m. This system is subjected 10 (a) Explain what is meant by the term velocity
to a suddenly applied force of 100 N. Write lag, as applied to control systems, and how
down the system differential equation and it may be reduced, (b) What is a three-term
hence determine (a) the initial acceleration, controller?
Control Principles 2 145
Q A 4.5 kg mass moves with a velocity of 8 m/s. If all of the kinetic energy stored by this mass is
transferred to a spring of stiffness coefficient 55 kN/m, calculate (a) the total displacement of the
spring, and (b) the value of capacitor that would be equivalent to this spring.
A
m 4.5 kg; v 8 m/s; k 55 103 N/m
(a) When this energy is transferred to the spring it will be stored as potential
energy because the spring will be displaced (compressed or stretched), and
1 1
then C farad
k 55 103
so C 18.2 µF Ans
A
F 2 104 N; k 2 105 Ns/m; R 3 105 Nm/s; t 0.5 s
ds
(a) FR ks newton
dt
ds
2 1 0 4 3 1 05 2 1 05 s
dt
ds
so, 2 30 20 s Anss
dt
146 Control Principles 2
R 300 kNm/s
F
20 kN
k 0.2 MN/m
Fig. 23
(b) Total displacement, S, occurs when the system comes to rest, i.e. ds/dt 0,
so putting this condition into the differential equation we have:
2 0 20 s
2
so, S 10 cm Ans
20
R 30
(c) System time constant, second
k 20
1.5 s
s S (1 et/ ) metre 10 (1 e0.5 /1.5 ) cm
s 2.83 cm Ans
A
F 120 N; m 15 kg; R 10 Ns/m
F1
m
F
R
F2
Fig. 7.24
d2 s
(a) The inertia force due to the mass, F1 m ms newton
dt 2
Control Principles 2 147
ds
The damping force due to friction, F2 R Rs newton
dt
and F F1 F2 newton
so, F ms Rs newton
120 15s 10 s
120 10 s
s
15 15
and, 8 s 0.67 s newton Ans
(b) Steady-state velocity occurs when the acceleration ceases, i.e. s 0, so
from the differential equation:
8 0 0.67 s
8
s metre/second 12 m/s Ans
0.67
(c) The electrical equivalents of mass, damping and force are inductance,
resistance and emf, respectively, so:
L 15 H; R 10 ; E 120 V
R L
10 Ω 15 H
E
120 V
Fig. 25
When the switch is closed at time t 0, the current will grow to some steady
value, which is limited only by the resistance of the circuit.
dq
I q amp
dt
120
so, q 12 A
10
10 5θ 6θ 4θ
2n 0.8
n 0.8 0.8944 rad/s
0.8944
fn hertz
2
fn 0.142 Hz Ans
R L C
59.6 mH 68 nF
1.5 kΩ
‘1’
E
24 V
Fig. 26
Control Principles 2 149
A
R 1500 ; L 0.0596 H; C 68 109 F; E 24 V
(a) This is a second-order system in which the charge (q) will increase from zero
to its final steady-state value, with an oscillatory component.
d2q dq q
EL 2
R volt
dt dt C
E d2q R dq q
L dt 2 L dt LC
24 d2q 1500 dq q
2
0.0596 dt 0.0596 dt 0.0596 68 109
d2q dq
402.68 2
25 167.8 2.467 108 q Ans
dt dt
(b) The general form of equation for a damped second-order system is:
Y d2 x dx
2 2n 2n x
a dt dt
2n 25 167.8
25 167.8
2 15 706.7
0.8 Ans
(d) fd fn 1 hertz
2
R
k
Fig. 27
A
m 25 kg; R 22 Ns/m; k 12 N/m
(a) F ms Rs ks newton
25s 22 s 12 s
F 22 12
s s s
25 25 25
F
s 0.88 s 0.48 s
25
Y
x 2n x 2n x and comparing coefficientts
a
2n 0.48
n 0.6928
0.6928
fn
2
fn 0.11 Hz Ans
(c) fd fn 1 2 hertz
0.11 1 0.6352
fn 0.085 Hz Ans